ring net-
um-noise
f dynamic
chosen in
quired. A
dynamic
ggle” and
uracy can
stereo in-
P.
O assess.
s contours
f a stereo
f the con-
he correct
precisely
it appears
in terms of
| is l or2
o clear the
»n, and ap-
to clear all
tive orient-
te in about
ive cycles.
t clears Y
erator by a
t the ability |
arallax and |
arge part of
y be so far
resulting X
tion circuits
that a model
reduces the
ably. A pro-
esa separate
ing scanning
the resulting
ime an angle
lel surface at
rn to resolve
JF; UM TO f
ivailable
small topographic structures should be improved
by this technique, since the pattern will tend
to conform more closely to the surface of the
model. A machine embodying this feature is now
under construction.
Height Ambiguity
The performance of Stereomat becomes
erratic whenever the model presents more than
one surface or level. Height ambiguity is most
troublesome during large-scale plotting, when
the ambiguity exceeds about 1 percent of the
flying height. Partial tree cover is the most
common example of this situation, particularly
when the trees are notin full leaf, and the ground
is visible between the branches. Tall buildings
present another example of height ambiguity
which is particularly serious when buildings of
various heights are clustered together within the
area of the scanning pattern.
Deterioration of machine performance from
this cause is of two types; (1) height errors in-
troduced by the vertical wandering of the pattern
between the limits of the ambiguity; (2) failure
of the correlation circuits to derive useful data
signals, thus disabling the machine completely.
The dual C.R.T. arrangement will probably
reduce the incidence of the latter effect, but at
the moment there is not a complete solution for
the problem of height errors.
Density of Transparencies
The plates for use with Stereomat on pro-
jection plotters should be of moderate contrast
and of relatively low density, in order to con-
serve the available scanning light. The density
should not exceed 1.0 at any point in the image,
when averaged over a circular area 1 millimetre
in diameter. The contrast of the plates should be
as great as possible, consistent with the above
density limit and with the preservation of detail
in highlight areas.
Plates for Stereomat should never be pre-
pared by means of a dodging printer or by an
unsharp masking technique, unless special pre-
cautions are observed. All dodging processes
remove the larger structure from image detail
and enhance fine structure and boundary defi-
nition. In the numerous areas of an image where
fine structure is absent, dodging removes the
only information available for stereo perception.
Plates may be printed on an electronic dodging
printer, if the scanning spot that exposes the
plate is defocussed to give an effective diameter
of about 5 millimetres (about 10 millimetres to
the eye). Plates printed on a fluorescent dodging
printer, or by an unsharp masking process, re-
quire a screen or mask separation of about 1
centimetre. The purpose of these precautions is
AUTOMATIC STEREO 11
to confine the dodging action to image detail
larger than that which is useful for parallax
sensing.
Special Purpose Plotters
The development of simplified Stereomat
instruments for special purposes is suggested.
An instrument for cross-sectioning, for example,
would provide X, Y, and Z read-out for coupling
to a computer but would not necessarily include
contouring circuits or XY transport facility.
A modified Stereomat could be coupled to
a computer to automate some engineering pro-
cedures such as route location. The computer
would issue commands to the stereo plotter, pro-
cess the readings, issue new commands, and so
on in a continuous optimi-ing sequence.
PREDICTIONS
The special skills required of the stereo-
plotting machine operator are hard to analyze and
define; perhaps because of the intimate mixture
of mechanical and interpretive activities in-
volved. Automatic stereo releases the operator
from the tedious mechanical functions of sensing
and clearing parallax, and thereby permits him
to concentrate on interpretive functions.
With human operation the intimate mixture
of mechanical and interpretive functions is not
a limitation. With automatic operation however,
stereo plotting divides naturally into a rapid
mechanical phase and a human interpretive phase.
Procedures may have to be adapted to permit
these phases to be performed at different times.
The drawing of planimetry from ortho-photographs
is an example of such an adaptation.
No doubt the need for human participation
and monitoring will be greatly reduced as ma-
chines and methods are improved and adjusted
to each other. No doubt also the pressure to
automate the remaining phases of map production
will increase, and techniques will be developed
to meet the demand. Eventually some of the
simpler interpretive tasks may yield to automation.
For example, the recognition of simple image
forms and the tracing of linear structures should
be possible using techniques similar to those
used for slope detection.
At present the Helava Analytical Plotter
appears to be the best basis for an automatic
stereo instrument. The development of a non-
mechanical plotter can be foreseen however,
and such an instrument could be expected to in-
crease plotting speeds even further.
The great capacity of automatic methods
will probably lead to a new disposition of photo-
grammetric and cartographic facilities. The
situation is in many ways analogous to that of
the digital computer, in that a single installation
" FIOLGOSGSUI UL DATO UL LUCA Al ch SS i chy fm sme ms ome om
Canada.
1ed,
sst
1at
1e
usly
f
lates
a of
the
of
b
atrol-
photo
b at
2e8Sary
ang
achine
spot
as
ns
at
for
efore
tions
a
will
lace
